Simulation of arctic low-level clouds observed during the FIRE Arctic Cloud Experiment using a ...

54 Pages·2000·0.34 MB·English

1
Simulation of arctic low-level clouds observed during the FIRE Arctic
Cloud Experiment using a new bulk microphysics scheme.
SHORT TITLE: Simulation of arctic low-level clouds
Eric Girard *and Judith A. Curry
Department of Aerospace Engineering Sciences
CB 429, University of Colorado, Boulder, CO 80309-0429
* Corresponding author
Submitted to J. Geophys. Res. on December 1, 1999, special section of the FIRE Arctic
Cloud Experiment. In revised version on April 10, 2000.
2
Abstract. A new bulk cloud microphysics scheme that accounts for aerosol microphysi-
cal properties and size distribution is implemented into the single-column version of the
ARCSyM. This scheme is distinguished from other bulk microphysics schemes by its
prognostic determination of cloud particle number concentration and saturation ratio.
The new scheme is compared to a simpler bulk microphysics scheme and observations
taken during the FIRE Arctic Cloud Experiment in May 1998. Qualitatively, the two micro-
physics schemes are generally in agreement with the observed cloud formation and evo-
lution. Comparison with aircraft measurements at three times shows that the new
scheme better discriminates cloud phase and reproduces reasonably well the observed
liquid and ice water content for two cases. The better performance of the new scheme is
attributed to its more elaborated treatment of the freezing process which is made possi-
ble by the prognostic determination of cloud particle number concentration and the
assumption of a bimodal lognormal cloud size distribution. Sensitivity studies are per-
formed to assess four aerosol microphysical properties on the evolution of cloud micro-
physical processes. Results show that the IFN concentration, the aerosol number
concentration, the slope of the aerosol size distribution and the aerosol solubility may
impact substantially on cloud phase and total water content. The liquid water path and
ice water path can vary by as much as 100 g m-2 locally as a result of the variation of
these parameters related to aerosols.
3
1. Introduction
The importance of arctic clouds and radiation to the regional and global climate is
summarized by Curry etal. [1996]. The difficulty in simulating and remotely sensing
these clouds motivated the FIRE Arctic Cloud Experiment [Curryetal., 2000], which was
conducted in conjunction with the Surface Heat Budget of the Arctic Ocean (SHEBA)
experiment [Perovichetal., 1999]. The difficulties in modelling arctic clouds have been
hypothesized byCurryetal.[1996] to arise from the complex vertical structure of the sta-
ble arctic atmosphere, the presence of mixed-phase clouds, and the susceptibility of the
arctic clouds to modification by aerosols.
Curryetal.[2000] describe an initial application of the SHEBA/FIRE dataset to
evaluating cloud parameterizations in the context of single-column model simulations.
Results of simulations for May 1998 showed that the models tend to underpredict low
cloud amount and the column liquid water path. The underestimation of the column liquid
water path was in part attributed to inaccurately representing mixed-phase clouds as
entirely crystalline.
Mixed-phase clouds occur at high frequency in the Arctic from November to June
[Curryetal., 1990; Pinto, 1998; Perovichetal., 1999]. From observations obtained dur-
ing autumn from the Beaufort Arctic Storms Experiment (BASE), Pinto[1998] hypothe-
sized that mixed-phase clouds are maintained in the arctic boundary layer through a
balance of condensation of liquid water through cooling and heterogeneous freezing of
the cloud drops. Jiangetal. [2000] simulated the observed case study described by
Pinto and found strong sensitivity of the phase of the cloud to the concentration of ice
forming nuclei (IFN). Simulations by GirardandBlanchet(GB)[2000a] with an explicit
4
aerosol-cloud microphysics model have shown that low-level clouds may remain in a
mixed-phase state as long as 10 hours in the Arctic during winter.
Bulk microphysical parameterizations used in climate models [e.g.Fowleretal.,
1996] typically distinguish the phase of cloud water by the cloud temperature. The dis-
crimination between the liquid and ice phases may be specified in terms of a single cutoff
temperature, or as a function of temperature with the ratio of ice water content to liquid
water content varying from 0 at 0oC to 1 at -40oC.Curryetal.[1996] cite observations of
liquid drops in arctic clouds at temperatures as low as -32oC, and clouds that are com-
pletely glaciated at -14oC. Clearly, a simple temperature discriminator is not sufficient to
diagnose the phase of supercooled clouds in the Arctic.
The complexity of the interactions between arctic cloud microphysics and aerosol
are discussed by GB[2000a, b]. Aerosol composition and concentration modify cloud
microphysical properties through their ability to nucleate ice and water. Blanchetand
Girard[1995] hypothesized that the microphysical properties associated to Arctic haze
aerosols in the Arctic may alter the cloud particle mean diameter and consequently
increase the precipitation efficiency. This would result in a larger dehydration rate of the
air mass which in turn would lead to a decrease of the downward longwave radiation flux
at the surface. Other indirect effects of aerosols such as the Twomey and the Albrecht
effects [Twomey, 1991; Albrecht, 1989] may strongly affect cloud microphysical proper-
ties.
Current bulk microphysics schemes used in general circulation models generally
describe cloud with four prognostic variables: liquid, ice, rain, and snow water content
[e.g.Fowleretal., 1996]. In these schemes, the effective radius of cloud particles is
5
parameterized as a function of temperature [e.g.OuandLiou, 1995] or liquid or ice water
content [e.g.McFarlaneetal., 1992]. Current moisture schemes are therefore of limited
utility in studying cloud microphysical properties changes resulting from aerosol forcing.
To account for the influence of aerosol on cloud microphysical properties, prognostic
equations must be included for either particle concentration or size, in addition to con-
densed water content.
In this paper, we describe a new bulk microphysics scheme that is capable of
accounting for the impact of aerosol on cloud nucleation. This scheme is developed by
merging the cloud schemes from GB[2000b] for cloud liquid and cloud ice and the pro-
duction of precipitation following Grelletal.[1995]. The new microphysics scheme pre-
dicts the ice and liquid water content, rain and snow, number concentration of cloud
droplets and ice crystals, and cloud saturation ratio. Aerosol concentration and composi-
tion are prescribed. This scheme is tested against observations taken during the FIRE
Arctic Cloud Experiment in May 1998. The new microphysics scheme is also compared
to the original version of the microphysics scheme developed by Grelletal.[1995] that
does not include prognostic equations for particle concentration or saturation ratio.
Finally, a sensitivity experiment is carried out to assess the effect on cloud formation of
changing the IFN and aerosol concentration, the slope of the aerosol size distribution
and the aerosol solubility.
2. Model Description
The single-column version of the Arctic Regional Climate System Model (ARC-
SyM) [Lynchetal., 1995] was used in our study. The ARCSyM is based upon the NCAR
6
regional climate model Version 2 (RegCM) [Giorgietal., 1993a,b]. The model has 38
vertical levels with 13 levels between 900 hPa and the surface that allows for a high res-
olution simulation of the vertical structure of the boundary layer. In this paper, a brief
description of ARCSyM physics package is given. The bulk microphysics schemes used
in this study are described in details thereafter. One is referred toLynchetal.[1995] for a
more extensive description of the other components of the model. In the ARCSyM, the
planetary boundary layer scheme of Holtslagetal.[1990] is used. The model uses the
CCM2 solar radiation scheme [Briegleb, 1992a,b] and the RRTM infrared radiation
scheme [Mlaweretal., 1997]. The cloud scheme ofGrellet al. [1995] is used for simulat-
ing cumulus cloud processes.
Two bulk microphysics schemes have been implemented in ARCSyM: (1) the
NCAR/MM5 Penn State cloud scheme [Grelletal., 1995], hereafter referred to as the
MM5 microphysics scheme, and (2) the new microphysics scheme developed in this
research. The MM5 cloud scheme is based on the bulk microphysics scheme of Grellet
al.[1995] and is an upgraded modified version of the microphysics scheme of Dudhia
[1989].Prognosticvariablesare:the icemixingratio,liquidmixingratio,snowmixingratio
and rain mixing ratio. Microphysical processes simulated include: nucleation of ice crys-
tals, nucleation of water droplets, condensation, evaporation, deposition, sublimation,
autoconversion of cloud droplet to rain, autoconversion of ice crystals to snow, melting/
freezing of snow/rain and of cloud droplets/ice crystals and collision processes.
The new scheme is distinguished from the MM5 cloud scheme by the addition of
three prognostic variables: the saturation ratio, the number concentration of cloud drop-
lets and ice crystals. These prognostic variables added to the ice and liquid mixing ratios
7
allow for determining the mean diameter of ice crystals and water droplets diagnostically.
Cloud particle size distribution is assumed to be a superposition of two lognormals with a
standard deviation of 1.3, each lognormal representing one phase, followingGB[2000b].
This choice of cloud particle size distribution is based on simulation of arctic low cloud
with an explicit microphysics model [GB, 2000a] and observations [Rogers and Yau,
1989]. Although observations indicate that the gamma size distribution can also be used
to represent cloud spectra in models [RogersandYau, 1989], the lognormal distribution
is preferred here since it is more representative of the cloud spectra obtained with the
explicit microphysics model [GB, 2000a] which has been used to develop the bulk micro-
physics scheme.
The new scheme and the MM5 scheme are based upon eight and five prognostic
equations respectively. The five common prognostic equations are:
water vapor mixing ratioq:
v
¶ q ¶ q ¶ q ¶ q
v = –u v –v v –w v –PCC –PCI –PII–PID+ PREC + PRES + d (1)
¶ t ¶ x ¶ y ¶ z qv
cloud liquid mixing ratioq:
c
¶ q ¶ q ¶ q ¶ q
c c c c
= –u – v –w +PCI + PCC + PRM –PCA –PRG –PRH
¶ t ¶ x ¶ y ¶ z
–PACCC –PRSC + d (2)
qc
cloud ice mixing ratioq:
i
8
¶ q ¶ q ¶ q ¶ q
i i i i
= –u – v –w +PII + PID+ PRG+ PRH – PRA– PACCI
¶ t ¶ x ¶ y ¶ z
–PRSI –PRM + d (3)
qi
rain mixing ratioq:
r
¶ q ¶ q ¶ q ¶ q
r = –u r – v r – w r –PREC –PKR+ PCA+ PACCC +d
¶ t ¶ x ¶ y ¶ z qr (4)
snow mixing ratioq:
s
¶ q ¶ q ¶ q ¶ q
s = –u s–v s–w s–PRES –PKS + PRA+ PACCI + d
¶ t ¶ x ¶ y ¶ z qs (5)
The new scheme additionally includes the following prognostic equations:
number concentration of cloud dropletsN:
c
¶ N ¶ N ¶ N ¶ N
c c c c
= – u –v – w
¶ t ¶ x ¶ y ¶ z
r
– ----------------(PCA+ PRSC +PACCC–PRG–PRH – PCI + PRM)+d (6)
m(D ) Nc
c
number concentration of ice crystalsN:
i
¶ N ¶ N ¶ N ¶ N
i i i i
= –u –v –w
¶ t ¶ x ¶ y ¶ z
r
– ---------------(PRA+ PRSI +PACCI–PRG–PRH – PII +PRM)+d (7)
m(D ) Ni
i
saturation ratioS:
dS –L ¶ T ¶ T ¶ T ¶ T p
------ = ------------ + u +v + w + -------[PCC + PCI + PII + PID– PRES –PREC] (8)
dt 2 ¶ t ¶ x ¶ y ¶ z e e
R T s
v
9
where u, v, w, p, T, r , L, e, R and d are respectively the zonal velocity, longitudinal
s v
velocity, vertical velocity, pressure, temperature, air density, latent heat of fusion, vapor
partial pressure at water saturation, gas constant for water vapor and the turbulent verti-
cal mixing. e =R/R where R is the gas constant for dry air. The terms m(D)represent
v d d
the mass of the ice crystal/cloud droplet of diameterD, which is the mean diameter of the
size distribution. The variables related to moist processes (PRES,PREC,PRA,PCA,
PRM,PID,PCC,PCI,PII,PACCI,PACCC,PRSC,PRSI,PKR,PKS,PRH,PRG) repre-
sents the sources and sinks of the microphysics fields and they are defined in appendix
A. Detailed formulation of these microphysical processes is given in Dudhia[1989] and
Grelletal.[1995]. The saturation budget equation follows the formulation of Rogersand
Yau[1989] (p.119). Srepresents the mean value or large scale value of the saturation
ratio of the grid cell. Subscale variations ofSare not considered. The first term in (8) rep-
resents the change of Sdue to air mass radiative and advective cooling and the change
of S due to condensation and deposition. In our simulations with the single column
model, the advective terms in equations (2) to (7) are neglected due to the absence of
observations. Advective tendencies in the equations for water vapor and saturation are
provided by the ECMWF reanalyses as described below. The vertical eddy diffusion (d )
is determined by the K-theory. The eddy diffusion coefficient is a function of the local
Richardson number as described inGrell et al. [1995]. Cloud entrainment is neglected.
Seven microphysical processes are treated differently in the new scheme relative
to the MM5 scheme: the cloud water droplet nucleation (PCI), ice crystal nucleation (PII),
the condensation/evaporation onto/of cloud water droplets (PCC), the deposition/subli-
mation onto/of ice crystals (PID), the heterogeneous freezing of water droplets (PRH),
10
the homogeneous freezing of water droplets and interstitial aerosols (PRG) and the auto-
conversion of ice crystals to snow (PRA). The treatment of these microphysical pro-
cesses is described in GB[2000b] for the new scheme and in Grelletal.[1995] for the
MM5 scheme. In this paper, due to their important role in the results in this research, the
treatment of the heterogeneous freezing of cloud droplets and autoconversion of cloud
ice to snow is described below.
2.1. Heterogeneous Freezing
The freezing of cloud water droplets is an important process that determines the
formation and longevity of mixed-phase clouds. Laboratory experiments [e.g. Prup-
pacherandKlett, 1997] and measurements in the Arctic [Pinto, 1998] have shown that
cloud glaciation varies as a function of the temperature and the cloud droplet size; the
largest having more chances to freeze due to their higher probability to encounter an IFN
in the atmosphere.
In the new bulk microphysics scheme, the heterogeneous freezing of cloud water
droplets depends on cloud droplet size, temperature and IFN concentration. The param-
eterization ofMeyersetal.[1992] is used for the IFN concentration. The median freezing
temperature (T ), which is defined as the temperature at which half of the cloud droplets
m
of a certain size freeze, is determined by the parameterization of Heverley [Pruppacher
and Klett,1997, p. 351] as follows:
1 (cid:230) (cid:230) aln2(cid:246) (cid:230) ¶ T(cid:246) (cid:246)
Tm = a--- lnŁ Ł --B----V----ł- + lnŁ ¶ tł ł (9)